We have had quite a ride. The eruption began unseen, on March 19. The new fissure opened on April 5, after the initial double cone had begun to wane. The new fissures sprouted a series of cones, mostly twinned. By May, all twins had exterminated one of the siblings, and the survivors had battled for supremacy leaving one winner. From now on, this would be a singular eruption.
On midnight May 2nd, everything changed. The tremor increased, the eruption went out, and then suddenly the world exploded. This kept happening, as often as 5 times per hour; 400-meter tall fire fountains were visible from Reykjavik. The seismographs remained extremely noisy for a long time, many weeks, while the volcano kept booming and the lava rose and fell around every boom and bust. Over time, the strength of the booms diminished and the eruption became a pulsing one, with a lava pulse (and a boiling lava lake) every 8 or 9 minutes. Amazingly, during all this time the eruption volume remained constant.
There was a general expectancy among volcanologists, volcano watchers and the tourist board that the eruption could continue for years. But on June 28 the eruption suddenly stopped, only to resume a few hours later. This has since become a pattern. The seismographs suddenly go silent and lava retreats out of sight, leaving an empty cone. Then the noise slowly increases, lava rises and a flood of boiling lava appears, looking like a lava tsunami coming down the sides of the cone. The flood diminishes, the seismographs goes flat and the cycle repeats. The duration of the active periods has varied from a few hours to a full day but apart from a lengthening there is no clear pattern. And yesterday the volcano flat-lined and did not come to life again for over a day. Is this the beginning of the end, or is it the end of the beginning? Who knows. GPS measurements show deflation around the eruption: lava is erupting faster than it is being replenished. That can not last forever. On the other hand the remarkably constant effusion rate shows that the eruption is not limited by available pressure but by the carrying capacity of the conduit.
In the mean time, the Iceland Meteorological Office ordered extensive hill fog to blanket the eruption into extinction. That didn’t work. Iceland’s engineers experimented with other types of eruption control. On May 14 they began work to build two walls, in order to contain the lava field. To our amazement it worked for a while but eventually the walls were overtopped. Other walls were build to protect Natthagakriki (June 15) and the coastal road (June 25). These still hold, helped by the fact that the lava gave up and decide to flow the other way. The new walls haven’t really been tested yet. If the eruption now ends, then the engineers will be happy and the government will declare that the battle against the Earth was won and take credit. If the eruption continues, then all bets are off.
The recent eruption has shown four distinct phases, with possibly a fifth just happening
- Constant effusion. This was the start when the lava flowed continuously;
- Intermittent fountaining. This was an exciting phase, approaching a strombolian eruption;
- Pulsing with a few minutes of activity followed by some 5 minutes of quiescence; each pulse produced spattering and mini-fountains, with quiet flows in between;
- Intermittent activity, with bubbly lava lakes for a time of several hours or longer with lava flooding, followed by a sudden retreat where the lava disappears and the seismographic first show pulses and then goes dead quiet. After some hours (or longer) the seismograph begins to show a bit of noise, which increases while the lava slowly rises up again. Over recent days the pulsing during the turn-off has become very weak.
- (Failure. The eruption interrupts but does not restart)
The amazing aspect is that the average flow rate did not change over this time, at least until the last few days. All measurements have returned 13m3, ever since early May, while the eruption went through these changes. Neither has the lava composition changed much, although there have been some minor variations. What is happening?
Let’s first look at the lava itself. It has a mantle-like composition, i.e. a form of basalt. Basalt may not be what you want in your plumbing, but in fact it flows quite well as long as it remains insulated. It has a low viscosity and is low in silicates (the two are related). The composition has shown that the magma formed at a high temperature of around 1250K at a depth of 15-20 km. It erupts at a bit lower temperature of 1170K. It contains some amount of CO2, and less SO2. This did not change during the changes of the eruption style. The dissolved water content has not been reported but for Iceland eruptions is typically around 0.8%.
The seismograph signal shows the noise that is generated as this magma flows up the conduit that connects to the surface. The signals are shown for high frequency (higher than 2Hz) and low frequency (less than 0.7Hz)
This is an example of the high frequency signal, showing the sudden stop.
Here is the low frequency signal, showing much less of a change.
What causes the noise? First, notice there are no sharp spikes visible. This means there is no rock cracking, which would show up as many small earthquakes. The lack of crackling shows that the conduit (and the dike underneath) is stable. The plumbing in this eruption is now well established and it is not in need of work. That will change only when the magma retreats from the conduit. We do see occasional rock falls on the steep inner side of the cone, giving a drawn-out signal lasting a minute or more. There are none in the plots shown here. Instead, the signal we see comes from magma moving through the conduit.
Magma can flow in two ways. The flow can be laminar, like honey creeping over a surface, or turbulent, like water in a steep river bed. Laminar flows are silent. The honey that is in touch with the surface is almost stationary, and the further it is from that surface the faster it flows. So you never have a fast flow directly over a corrugated surface. Turbulent flows are very different. The liquid is moving at different speeds and even in different directions and there is a lot of interaction between fast flows and surfaces. This seems the noise that we see. When the seismograph is noisy, something is causing turbulence in the flow. When it flat lines, the flow is undisturbed and laminar – or it has stopped.
Do be aware that the seismographs can pick up noise from other sources. Especially the low frequencies pick up movement over large areas, sometimes including visitors. (They also see large earthquakes across the entire world.) Wind can affect both plots: the plot thickens and becomes noisy. When it storms, eruptions become hard to see.
The signal does not tell us where the flow noise is located. It could be in a deep conduit, it could be on the surface or in a lava tube. If you look hard at the high frequency signal, squint a bit, and use a bit of imagination, you can see a hint of pulsing just before the end, lasting maybe 15 minutes with pulsing repeating over 3 ot 4 minutes. That can only be in the conduit, so I am assuming that the entire signal comes from the vertical pipe through which the magma rises to the surface.
What causes a flow to become laminar or turbulent? The smoothness of the surface is important. A river can be beautifully laminar where it is wide and has a smooth sandy floor, but turbulent where it becomes rocky or changes its width. The viscosity is also important. A fluid with high viscosity (internal stickiness or friction: think honey) tends to flow laminar, while a low viscosity fluid (such as water) very quickly becomes turbulent.
The magma in Fagradalsfjall is a type of basalt which has low viscosity. That is both because of its composition (it has few silicate crystals which easily stick together) and because of its high temperature. The lava channels show a fast flow, and this is indicative of a low viscosity lava. By the time it gets to the end points in Meradalir and Natthagi, it has cooled down and when flowing on the surface it behaves more viscous, although not nearly as much as rhyolite which just refuses to flow and sticks to the ground.
We would therefore expect that our magma can easily become turbulent. It doesn’t in the lava channels: even though the lava flows fast, it is still laminar. That is by and large also true underground, as long as it flows through wide pipes or tubes. Indeed, the seismographs was noisy when magma was still breaking through to the surface and did not yet flow, but they became rather quiet when the eruption was well established. The flow underground was also laminar. But later the eruption became fountainous and after that bubbly. And the noise went through the roof. There was turbulence in the magma. The fact that this happened while the lava was bubbling suggest that there was gas in the plumbing. The gas in the low viscosity magma caused turbulence. Where did the gas come from?
Let’s take a step back, or rather down. The magma rises up because it is buoyant: it has lower density than the surrounding rock. The hotter it is, the lower the density, and so hotter lava (if there is a choice) rises faster. As it rises, the depth becomes less and there is now less pressure from the weight of the rock above. The pressure in the magma decreases. At the same the temperature also drops a bit. To give some numbers, at 15-20 km depth where the magma was sourced, the pressure was around 400 Mpa (4 kbar if you prefer) and the temperature was around 1220 K. (The melt had actually formed even deeper, perhaps 25 km.) By the time it entered the dike, at 6 km depth, the pressure was down to 150MPa and the temperature around 1200 K, and at the point where the dike connected to the conduit, perhaps 2 km depth, they were 50 MPa and 1190 K respectively. The magma erupted at the surface with a temperature of 1170 K.
The origin of the gas lies in the changing conditions during the rise. Liquid magma can contain a limited amount of volatiles, such as water and CO2. If there is more of these then can dissolve into the magma, the excess is expelled and becomes a gas – a vapour inside the magma. The maximum amount that can dissolve in the liquid is called the solubility. It is different for each volatile. To give a rough number, basalt at 1200 K and a pressure of 50MPa can contain around 2% (by weight) of water. This amount decreases rapidly with pressure: by the time the pressure is down to 5 MPa (200 meters depth), the solubility is down to around 0.5%. It scales roughly as the square root of the pressure.
Temperature has the opposite effect: as the magma cools it can contain more water. You can see this effect when heating water in a pan. As the temperature rises, bubbles appear in the water. This is gas coming out of the liquid. Let the water cool and the bubble disappear again, taken up by the water. But in this magma the effect of temperature is fairly minor. The solubility of water scales roughly as 1/T (with the temperature T in Kelvin), and the temperature drops by only around 5% between 15 km depth and the surface. The pressure reduces much more dramatically, and it wins the battle. So while the magma rises, it tries to dry out and expel excess water. You may want to think of a volcanic eruption as a giant dryer.
Icelandic magma is pretty dry to begin with, but not that dry. At 0.8%, the Fagradalsfjall magma reaches a problem at 500 meters. At that depth it becomes saturated. As it rises further the magma begins to expel water and water vapour (steam) develops in the magma. The magma becomes gassy, and just like a human body after a good meal, it becomes windy and noisy.
CO2 goes through the same process as water but it does so at much greater depth. Mantle plumes may contain 1 % CO2 by weight, but this already turns into gas at a depth below 5 km. Some of this CO2 finds its own way to the surface and some remains as a gas inside the magma. By the time basaltic magma is at 1 km depth there is little dissolved CO2 left.
So water in Fagradalsfjall’s magma produces vapour during the last 500 meters of the ascend. This is not an easy process. A phase change (liquid to vapour or liquid to solid) needs something to hold on to. Pure water can in fact be cooled to well below freezing while still staying liquid. But shake it a bit and it freezes instantly. The water was supercooled. Air too can be supersaturated whilst not producing clouds . But when a passing plane disturbs it, instantly a contrail forms. It is the same with magma: it can become supersaturated with water but still reluctant to let it go. It takes time for the water to evaporate out of the liquid. If this is longer than the time it takes the magma to reach the surface, then the water will stay in the magma as an unwelcome passenger.
(I remember a camping trip (in Africa!) when after a chilly night we tried to pour water from a bottle into a cup. It froze on the way, mid-air. The water had become supercooled.)
When the water turns to gas, it forms small bubbles inside the magma in a process called bubble nucleation. Initially these are tiny, microscopic even. Nucleation is much easier when there are crystals in the magma: they provide a surface on which the bubbles can grow with ease. If there are no crystals, bubbles form with difficulty and the magma becomes supersaturated. But take supersaturated magma and add nucleation sites (crystals) and bubbles instantly form everywhere. If a magma rises rapidly, it will become supersaturated because the water has no time to respond to the decompression. But as the pressure continues to fall, at some point the supersaturation may become so high that nucleation accelerates anyway. Suddenly, gas is everywhere. The magma becomes fizzy and turbulent.
The bubbles are very buoyant and try to rise. But the magma is too viscous for that. It is worst for the smallest bubbles: friction with the magma locks them in place. Larger bubbles find it easier to rise, especially in a low viscosity magma. Let’s assume that the magma in the conduit rises at a speed of 1 m/s. That is a reasonable value for Fagradalsfjall: it gives the right flow rate (13 m3/s) for a conduit that is 4 meters across. The bubbles will move up a little bit faster, but not much faster. Even in Fagradalfjall they will only go faster by a few cm/s. The magma now becomes a mix of liquid and bubbles.
As more of the water becomes gas, the bubbles grow and take up a larger fraction of the volume. As the bubbles become mobile, they collide and can merge, or take in more water from the surrounding magma. And they also expand because the pressure is dropping as the magma rises. The bubbles can grow as large as a few centimeters. So as the magma approaches the surface, more and more of the volume is taken up by gas.
When bubbles take up more than half the volume, the bubbles merge into gas pockets. These are called ‘slugs’ and they take up the full diameter of the conduit, pushing the magma out of the way. If 90% of the volume is gas, then the slugs merge into a column of gas and magma is pushed to the side, but this may not happen in real volcanoes. In a bubble flow, the bubbles are stuck and rise with the magma. But in a slug flow, the slugs can rise rapidly because of their low density and because their smooth surface pushes the magma out of the way. The bubbly flow is sluggish and the slug flow is not.
What kind of speed can we reach? Here, the change of density is important. If half the volume is taken up with bubbles, the density of the mixture has halved. The magma now becomes very buoyant. If we start at 500 meters depth and use all the energy in the buoyancy compared to the surrounding rock to accelerate, by the time we reach the surface the velocity can reach 100 m/s. At that speed, a ballistic trajectory can reach 500 meters height. This is about what the highest fountains in early May reached. (The reality is of course much more complex. Much of the energy is lost in friction in the conduit and the slugs don’t travel anywhere near that fast. On the other hand the slugs still expand and this expansion greatly adds to the velocity.)
Slug flows are the dominant cause of strombolian eruptions. Each slug, when arriving at the surface, causes an explosion both because of its speed and because the slug expands fast in the low pressure around it. It throws out the surrounding magma (lava?) with it; it fountains, fragments and falls. If there is debris plug on top of the conduit and/or stagnant lava, the slug can become more explosive, and produce ash. If the conduit is open the fragments are ballistic lava. Fargradalsfjall always had an open conduit.
This degassing of the magma was the driving force during the fountaining phase, and the eruption changed because it began to degas much more. Originally, when the eruption first began, the magma did little degassing: this was the time of the constant outflow which we saw coming from the first cone, and later from the fissure. The magma at this time may have been less supersaturated, so that the bubbles formed slower and never merged into slugs.
There can be several possible causes for the change to slug formation. The magma may have changed, and had a higher supersaturation. This was also the time that the flow rate increased to its current value of 13 m3/s: this is possible with the same conduit if the density or viscosity of the magma became a bit less. The change allowed for faster bubble nucleation.
The eruption became extremely noisy at this time: because of rapid degassing the flow became bubbly on the ascend and therefore turbulent. The whole conduit, from 500 meter down to the surface degassed together. I envisage this as starting near the top. The sudden appearance of many bubbles drives out magma, and this reduces the pressure lower down where the magma now carries less weight. This decompression increases the supersaturation and allows bubbles to form here, and so on. A bubble formation front accelerates downward and the whole column turns first fizzy and then bubbly, before the rising bubbles begin to form slugs. It is just like opening an overpressured bottle of carbonated water.
(No slugs were harmed (or produced) in the making of these movies)
(An alternative idea is that the seismograph noise that we see comes from the bubbles themselves, as they implode, explode and merge, so that the turbulence is not in the flow but in the bubbles.)
Why the 10 minutes between the strombolian fountains? At a speed of 1 m/s, it takes ten minutes for the 500 meters of degassed magma to be replaced by the gas-rich magma from below, after which the process could repeat.
The pulsing that happened later was different. There were no high fountains, and there was no large acceleration. There were no slugs. Also, the seismographic noise happened mainly during the pulse whilst in between pulses the flow was much less noisy. The magma now was less supersaturated and fewer bubbles formed. The same process as before happened but the bubbles never reached the slug threshold of 50% by volume. The magma became bubbly but not sluggy. The bubbly flow still reduced the density of the magma, and this caused the level of the lava at the top to go up. The lava lake filled up and overflowed. The lava degassed over a couple of minutes as the gas reached the surface. This caused the apparent boiling of the lake. Once the gas became depleted, the density of the lava became higher and the level of the lake went down again. The lower noise of the seismograph now showed much less turbulence from bubbles. Over the next 5-10 minutes the magma in the conduit was replaced by fresh magma, and the degassing would start again.
This process explains why the flow rate of the lava did not change. The rate at which magma ascended from below remained constant all through these phases.
The current phase of long quiescent and active periods is different again. The fact that the seismographs are flat during the quiet period suggests that the flow becomes laminar without bubbles. There is no gas: the magma is not supersaturated. The density is therefore also higher and the level of the lava lower. We don’t know whether there still is any lava flow during these periods but it is possible it still flows -silently- through a deeper tube towards Meradalir, below the layer of the rubble that we can see in the cone. However, it is also possible that the magma in the conduit stops flowing and the eruption interrupts. It depends on how buoyant the magma below the conduit is.
Slowly the low frequency noise increases. A bit of degassing is beginning but the volume of the gas remains small. As the degassing increases the density of the conduit decreases, but only a little because there is less water available for degassing. The magma begins to rise, and the pressure below begins to drop. This causes some supersaturation and slowly more gas comes out of the magma and forms bubbles. The density decreases as before, the lava level rises and the cone overflows. The process is slow enough (several hours) that the magma is always in equilibrium, able to shed the excess water and avoiding supersaturation. The new situation with the magma a little bubbly but not very much is stable: it will continue as long as nothing disturbs it. But a rock fall at the top or a slowing of the overflow will increase the weight below, and the bubbles begin to dissolve into the magma. There is a little bit of pulsing at this time, but with a shorter period of a few minutes. This suggests that the gas formed only higher up in the conduit. This is of course what you would expect if the is less water in the magma: the depth at which it become supersaturated becomes less. Once the gas is gone, the situation is stable again.
The volcano therefore can have two very different modes which are both stable: one with fizz and one without.The change from one to the other is unpredictable. In physics, this can give rise to a chaotic system with periods of constant behaviour followed by a random large change. If you are interested, look up the Lorentz attractor.
The changing eruption does not necessarily mean that the flow rate has decreased. However, the supersaturation of the magma is changing. Perhaps there is less water. There may also be a more prosaic reason. The eruption has added a lot of weight to the area. The pressure in the conduit is therefore a bit higher than before, and this can reduce the supersaturation.
What will happen next? The current situations is unlikely to last long. The long periods without lava suggests the eruption is on the edge, and could easily stop completely. If that happens (not unlikely), the conduit may block, the walls will survive and the government will declare victory and call elections. But if there is still flow from the mantle, then the pressure from below will increase again and the magma will look for another way out. In that case, after a while the rock-breaking earthquakes will restart and the eruption may eventually resume in a new location.
This may well be the end. It may also be the start of something new.
Albert, July 2021